Mono-, Di- and Tetra-iron Complexes with Selenium or Sulphur Functionalized Vinyliminium Ligands: Synthesis, Structural Characterization and Antiproliferative Activity.

A series of diiron/tetrairon compounds containing a S- or a Se-function (2a-d, 4a-d, 5a-b, 6), and the monoiron [FeCp(CO){SeC1(NMe2)C2HC3(Me)}] (3) were prepared from the diiron μ-vinyliminium precursors [Fe2Cp2(CO)( μ-CO){ μ-η1: η3-C3(R')C2HC1N(Me)(R)}]CF3SO3 (R = R' = Me, 1a; R = 2,6-C6H3Me2 = Xyl, R' = Ph, 1b; R = Xyl, R' = CH2OH, 1c), via treatment with S8 or gray selenium. The new compounds were characterized by elemental analysis, IR and multinuclear NMR spectroscopy, and structural aspects were further elucidated by DFT calculations. The unprecedented metallacyclic structure of 3 was ascertained by single crystal X-ray diffraction. The air-stable compounds (3, 4a-d, 5a-b, 6) display fair to good stability in aqueous media, and thus were assessed for their cytotoxic activity towards A2780, A2780cisR, and HEK-293 cell lines. Cyclic voltammetry, ROS production and NADH oxidation studies were carried out on selected compounds to give insights into their mode of action.


Synthesis and Characterization of Compounds
Compounds 2a and 2c [48], 4c [49] and 6 [50] were previously reported, whereas 2b, 2d, 3, 4a, 4b, 4d, 5a, 5b, and 6 are novel (Scheme 2). Once isolated, 2a-d slowly decompose in contact with air, whereas 3-6 resulted indefinitely air-stable. The sodium hydride(methoxide)-promoted dehydrogenative chalcogenylation of 1a-c, as described previously [48], provides access to the zwitterionic complexes 2a-d, in 60%-80% yields. This formal [C 2 H] + /C 2 E substitution (E = S, Se) presumably proceeds through the initial single-electron reduction of the cationic part of 1a-c. Consistent with this hypothesis, the monoiron complex 3, maintaining the C 2 -H unit, is a side product of the reaction leading to 2a, and may be viewed as the result of selenium incorporation along a fragmentation process initiated by electron transfer to 1a [51]. The chalcogenido moiety in 2a-d is readily oxidized with I 2 to the dimeric iodide salts 4a-d, containing an E-E bridge (77%-93%) [48]. Electrophilic methylation of 2a-d affords 5a-b (76%-86%). Instead, 6 is directly derived from 1a (80%), trapping the [SPh] fragment along the reaction of 1a with NaH [49]. According to combined X-ray diffraction analysis and NMR spectroscopy studies, the previously reported 2c and related R = Xyl containing complexes exist both in solution and in the solid state in the Z form, i.e., displaying the bulky xylyl group far from the chalcogen atom [48]. The salient NMR spectroscopic features of the new compounds, 2b and 2d, are in good agreement with those of 2c and analogues, thus indicating a Z configuration. For instance, the Cp rings and the methyl groups in the respective 1 [47], at 4.59, 4.55 (Cp), 3.69 (Me) and 2.65, 2.16 ppm (Xyl). The latter complex differs from 2b in the presence of a 4-tolyl substituent in the place of Ph, and its structure was confirmed by X-ray diffraction.
DFT calculations confirm that the Z isomers of 2b and 2d are more stable than the E form by about 6 kcal mol -1 (Figures 1 and 2). A comparison of computed bond lengths and angles indicates only small changes on replacing sulphur (2b) with selenium (2d). The most affected distance is Fe(2)-C(2), being 2.143 Å in 2b and 2.119 Å in 2d (CPCM/B97X calculations). The similarity between 2b and 2d is confirmed by the Mulliken population analysis, providing close values of partial charge for the μ-vinyliminium ligand in the two compounds. The higher stability of the Z isomers can be explained on the basis of the lower electrostatic repulsion between the chalcogen atom and the xylyl ring, as observable for instance in Figure S1 (Supporting information), where the electrostatic potential surfaces of E-2b and Z-2b are compared. According to combined X-ray diffraction analysis and NMR spectroscopy studies, the previously reported 2c and related R = Xyl containing complexes exist both in solution and in the solid state in the Z form, i.e., displaying the bulky xylyl group far from the chalcogen atom [48]. The salient NMR spectroscopic features of the new compounds, 2b and 2d, are in good agreement with those of 2c and analogues, thus indicating a Z configuration. For instance, the Cp rings and the methyl groups in the respective 1  The latter complex differs from 2b in the presence of a 4-tolyl substituent in the place of Ph, and its structure was confirmed by X-ray diffraction.
In both 2b and 2d, the HOMO is localized on a p-type orbital of E (E = S, Se) and to a lesser extent on C 3 , C 2 and on the iron centers (see Figures 1 and 2): this explains why the chalcogen atom E is the most probable site for molecular oxidation or electrophilic attack. The HOMO of 2d is located 0.21 eV higher than in 2b, so I2-oxidation of 2d to 4d is 6.5 kcal mol -1 more favorable than the  The 1 H and 13 C NMR spectra of 3 (acetone-d 6 solution) display two resonances for the N-bound methyls, in accordance with the iminium description of the [C 1 -NMe 2 ] moiety. Signals attributable to the C 1 , C 2 and C 3 carbons are observed at 218.6, 137.4 and 199.3 ppm, respectively, while the selenium centre is observed at 285.7 ppm in the 77 Se NMR spectrum.
In both 2b and 2d, the HOMO is localized on a p-type orbital of E (E = S, Se) and to a lesser extent on C 3 , C 2 and on the iron centers (see Figures 1 and 2): this explains why the chalcogen atom E is the most probable site for molecular oxidation or electrophilic attack. The HOMO of 2d is located 0.21 eV higher than in 2b, so I 2 -oxidation of 2d to 4d is 6.5 kcal mol −1 more favorable than the analogous reaction for 2b. Presumably, 4b and 4d containing xylyl groups, maintain the Z configuration of the N-substituents adopted in their precursors 2b and 2d [48]. Indeed, the NMR spectra of 4a-d suggest the presence of a single species in solution. On going from 2d to 4d, the Se center undergoes significant deshielding in the 77 Se NMR spectrum (from 282.5 to 556.4 ppm). DFT-optimized structures of 4b and 4d are shown in Figures S2 and S3.

Electrochemistry
Compounds 4c and 5a were selected for electrochemical characterization in acetonitrile, which was extended to the respective precursors 1c and 2c, and also to 1a. The main results are summarized in Table 2, and all cyclic voltammetric profiles (referred to the ferrocenium/ferrocene redox couple) are provided in the Supporting Information ( Figures S4-S14). In general, the investigated complexes exhibit one electrochemical reduction process, which occurs reversibly for 1a and 5a on the time scale of the experiment, respectively at −1.37 V and −1.29 V. However, the reduction observed for 1a is complicated by either two different processes occurring at very similar potentials, or a single process occurring in two steps ( Figure S6). As a consequence, a slightly high peak-to-peak separation (∆E p ) of 108 mV was recognized. Furthermore, 1a displays an irreversible oxidation at +0.65 V, whereas in the case of 5a several irreversible oxidation reactions were detected in the potential range from −0.44 V to +0.66 V.
As discussed above, the chalcogenido moiety of 2c can be chemically oxidized to 4c (Scheme 2), and the same conversion was investigated using electrochemical techniques. As expected, the cyclic voltammogram (CV) of 2c shows an irreversible oxidation at +0.12 V, ascribable to the generation of the cationic part of 4c. Correspondingly, the CV profile of 4c shows an irreversible reduction at −0.78 V, that could be assigned to the formation of 2c [61]. Further considerations are prevented due to the presence of iodide as the counteranion in 4c, which is redox active and leads to the deposition of degradation products on the surface of the working electrode. Table 2. Overview of the main oxidation and reduction potentials (V vs. Fc + /Fc) at a scan rate of 100 mV/s determined by cyclic voltammetry in MeCN for selected iron complexes. The peak-to-peak separation (∆E p ) is determined by the difference between two peak potentials for a given redox couple. a E pa for an irreversible process.

Compound
Oxidation

Cytotoxicity Studies and Stability in Aqueous Media
The air sensitive compounds 2a-d were excluded from the biological tests. The remaining compounds were preliminarily evaluated for their stability in aqueous media (data summarized in Table 6). The ionic compounds 4a-d, 5a-b and 6, which are slightly soluble in water, and 3 did not undergo significant modification in DMSO-d 6 /D 2 O solution after 72 h at ca. 37 • C, according to 1 H NMR spectroscopy (see Experimental for details). Approximately 50% degradation of 4b to unidentified species was detected after a further 72 h following addition of NaCl to the solution, whereas 4a,c,d did not change under the same conditions. IR spectroscopy was used to estimate the stability of 4a-d, 5a and 6 in contact with cell culture medium at 37 • C. Compounds 4b, 4d, 5a, and 6 remained intact after 72 h, whereas 4a and 4c were recovered at the end of the experiment together with other carbonyl species. Indeed, a significant amount of 2c was detected to be produced from 4c.
Three tetrairon complexes, i.e., 4b-4d, containing a S-S or a Se-Se bridge, and the diiron vinyliminium complexes 5b and 6, bearing a thioether function, possess potent cytotoxicity against the cancer cell lines, with IC 50 values in the low micromolar/nanomolar range. In particular, the activity of 4b, 4d, and 5d is superior than that of cisplatin and appears to overcome resistance issues, since comparable IC 50 values were determined on the A2780 and A2780cisR cell lines. However, selectivity is not observed compared to the HEK-293 cell line, apart from a moderate selectivity shown by 5a. The introduction of a Se-Se bridge on 1a leads to a dramatic decrease in activity, the diselenide derivative 4a being inactive towards all the investigated cell lines. In general, the strongest cytotoxic effect promoted by 4b,d, compared to 4a,c, reflects the higher stability in aqueous media of the former respect to the letter (see above). Table 3. IC 50 values (µM) determined for compounds 3, 4a-d (and their vinyliminium precursors 1a-c), 5a-b, 6, cisplatin and RAPTA-C on human ovarian carcinoma (A2780), human ovarian carcinoma cisplatin resistant (A2780CisR) and human embryonic kidney (HEK-293) cell lines after 72 h exposure. Values are given as the mean ± SD. a See reference [39].

ROS Production and NADH Oxidation
We previously hypothesized that the cytotoxicity of diiron vinyliminium compounds, 1, is mainly ascribable to redox mechanisms (see Introduction). As suggested by the DFT outcomes, electrochemical investigations and stability data (see above), the tetrairon-bis-cationic complexes 4a-d are susceptible to relatively facile reduction due to feasible disulphide(diselenide) to sulphide(selenide) conversion.
Even the reduction of the selenido-vinyliminium 5a appears slightly more favorable with respect to analogous non-functionalized vinyliminium complexes (Table 2) [63]. Therefore, the cytotoxicity of the S-and Se-derivatives, and especially 4b-d, is expected to involve interference of cellular redox processes. In order to investigate this aspect, we assessed the production of intracellular ROS levels induced by a selection of complexes (fluorescence measurements, using the peroxide-sensitive probe DCFH-DA). Thus, A2780 cells were continuously exposed to 4a, 4b, 4c, 5a, cisplatin (as a reference compound) and H 2 O 2 (as a positive control). Treatment with 4b and 4c showed a significant increase in the level of ROS after ca. 20 h of treatment with respect to the positive control ( Figure 4). Instead, 4a and 5a stimulated a ROS production close to that recorded for the basal levels; moreover, 4a did not show a significant effect even at higher concentration (100 µM). The marked difference in behavior between 4a (non cytotoxic) and 4b-d indicates that the stimulation of ROS production could be indeed a privileged way of antiproliferative action for 4b-d.
Molecules 2020, 25, x FOR PEER REVIEW 9 of 21 exposed to 4a, 4b, 4c, 5a, cisplatin (as a reference compound) and H2O2 (as a positive control). Treatment with 4b and 4c showed a significant increase in the level of ROS after ca. 20 h of treatment with respect to the positive control ( Figure 4). Instead, 4a and 5a stimulated a ROS production close to that recorded for the basal levels; moreover, 4a did not show a significant effect even at higher concentration (100 µ M). The marked difference in behavior between 4a (non cytotoxic) and 4b-d indicates that the stimulation of ROS production could be indeed a privileged way of antiproliferative action for 4b-d. In order to further evaluate the ability of compounds to interfere with physiological redox processes, we determined the catalytic activity of 4a, 4c, 4d, 5a, and 6 in the aerobic oxidation of NADH, using a previously documented UV-Vis method (Table 4) [64]. Indeed, nicotinamide adenine dinucleotide (NAD + ) and its reduced form (NADH) are important cofactors contributing to the maintenance of redox balance in cells [65], and the alteration of the NADH/NAD + ratio has been implicated in the anticancer activity of various late transition metal complexes [64,66]. Cationic diiron vinyliminium compounds without chalcogen-functions, i.e., 1a and 1c, were also included in this study for comparison, together with FeSO4 as a reference compound. All tetrairon compounds 4a, 4c, 4d displayed a moderate catalytic activity on NADH oxidation, comparable (or slightly superior) to that of their diiron precursors (1a, 1c). Surprisingly, the diiron compounds 5a and 6, featuring selenoether/thioether moieties, were able to retard the oxidation of NADH with respect to the blank experiment. Notably, TONs at 25 h were significantly lower for 5a and 6 than for FeSO4, the latter exhibiting no catalytic activity.  In order to further evaluate the ability of compounds to interfere with physiological redox processes, we determined the catalytic activity of 4a, 4c, 4d, 5a, and 6 in the aerobic oxidation of NADH, using a previously documented UV-Vis method (Table 4) [64]. Indeed, nicotinamide adenine dinucleotide (NAD + ) and its reduced form (NADH) are important cofactors contributing to the maintenance of redox balance in cells [65], and the alteration of the NADH/NAD + ratio has been implicated in the anticancer activity of various late transition metal complexes [64,66]. Cationic diiron vinyliminium compounds without chalcogen-functions, i.e., 1a and 1c, were also included in this study for comparison, together with FeSO 4 as a reference compound. All tetrairon compounds 4a, 4c, 4d displayed a moderate catalytic activity on NADH oxidation, comparable (or slightly superior) to that of their diiron precursors (1a, 1c). Surprisingly, the diiron compounds 5a and 6, featuring selenoether/thioether moieties, were able to retard the oxidation of NADH with respect to the blank experiment. Notably, TONs at 25 h were significantly lower for 5a and 6 than for FeSO 4 , the latter exhibiting no catalytic activity.

Synthetic Procedures and Compound Characterization
General details. The preparation, purification and isolation of compounds were carried out under a N 2 atmosphere using standard Schlenk techniques; once obtained, 3-6 were stored in air and 2a-d were stored under N 2 . Solvents were purchased from Merck and distilled before use under N 2 from appropriate drying agents. Organic reactants (TCI Europe or Merck) were commercial products of the highest purity available. Compounds 1a-e [39,42], 2a,c [48], 4c [49], and 6 [50] were prepared according to published procedures. Chromatography separations were carried out on columns of deactivated alumina (Merck, 4% w/w water). Infrared spectra of solutions were recorded on a Perkin Elmer Spectrum 100 FT-IR spectrometer with a CaF 2 liquid transmission cell (2300-1500 cm −1 range); IR spectra were processed with Spectragryph software [67]. NMR spectra were recorded at 298 K on a Bruker Avance II DRX400 instrument equipped with a BBFO broadband probe. Chemical shifts (expressed in parts per million) are referenced to the residual solvent peaks ( 1 H, 13 C) [68], or to external standard ( 77 Se, SeMe 2 ). 1 H and 13 C NMR spectra were assigned with the assistance of 1 H-13 C (gs-HSQC and gs-HMBC) correlation experiments [69]. Elemental analyses were performed on a Vario MICRO cube instrument (Elementar).
Compound 2b was prepared using the procedure reported in the literature for 2a and 2c [48], and a slight modification of the procedure was employed for 2d.

Synthetic Procedures and Compound Characterization
General details. The preparation, purification and isolation of compounds were carried out under a N2 atmosphere using standard Schlenk techniques; once obtained, 3-6 were stored in air and 2a-d were stored under N2. Solvents were purchased from Merck and distilled before use under N2 from appropriate drying agents. Organic reactants (TCI Europe or Merck) were commercial products of the highest purity available. Compounds 1a-e [39,42], 2a,c [48], 4c [49], and 6 [50] were prepared according to published procedures. Chromatography separations were carried out on columns of deactivated alumina (Merck, 4% w/w water). Infrared spectra of solutions were recorded on a Perkin Elmer Spectrum 100 FT-IR spectrometer with a CaF2 liquid transmission cell (2300-1500 cm −1 range); IR spectra were processed with Spectragryph software [67]. NMR spectra were recorded at 298 K on a Bruker Avance II DRX400 instrument equipped with a BBFO broadband probe. Chemical shifts (expressed in parts per million) are referenced to the residual solvent peaks ( 1 H, 13 C) [68], or to external standard ( 77 Se, SeMe2). 1 H and 13 C NMR spectra were assigned with the assistance of 1 H-13 C (gs-HSQC and gs-HMBC) correlation experiments [69]. Elemental analyses were performed on a Vario MICRO cube instrument (Elementar).

Chart 3. Structure of 3.
The reaction mixture for the synthesis of 2a was obtained as described in the literature, from 1a, gray selenium and NaH [48]. This mixture was filtered through a short alumina pad using THF as eluent, then the volatiles were removed under vacuum. Subsequent alumina chromatography of the residue led to isolate a red fraction using neat diethyl ether as eluent, corresponding to 3. A solution of 1b (180 mg, 0.259 mmol) in THF (15 mL) was treated with gray Se (200 mg, 2.53 mmol) followed by NaOMe (35 mg, 0.648 mmol). The mixture was allowed to stir at room temperature for 1 h, then it was filtered through a short alumina pad, using neat THF as eluent, under protection from air. The filtrate was dried under vacuum. The resulting residue was dissolved in CH 2 Cl 2 and the solution was charged on alumina. Elution with CH 2 Cl 2 removed the impurities and a green band was collected using THF as eluent. The title compound was isolated as a brown solid upon removal of the solvent under vacuum. Yield 129 mg, 80%. Anal. calcd. for C 30

Chart 3. Structure of 3.
The reaction mixture for the synthesis of 2a was obtained as described in the literature, from 1a, gray selenium and NaH [48]. This mixture was filtered through a short alumina pad using THF as eluent, then the volatiles were removed under vacuum. Subsequent alumina chromatography of the residue led to isolate a red fraction using neat diethyl ether as eluent, corresponding to 3.  The reaction mixture for the synthesis of 2a was obtained as described in the literature, from 1a, gray selenium and NaH [48]. This mixture was filtered through a short alumina pad using THF as eluent, then the volatiles were removed under vacuum. Subsequent alumina chromatography of the residue led to isolate a red fraction using neat diethyl ether as eluent, corresponding to 3. The title products were prepared using the procedure reported in the literature for 4c [49]. The title products were prepared using the procedure reported in the literature for 4c [49].

Synthesis of [Fe2Cp2(CO)(μ-CO){μ-η 1 :η 3 -C 3 (R')C 2 (EMe)C 1 N(Me)(R)}]I (R = R' = Me, E = Se, 5a; R = Xyl, R' = Ph, E = S, 5b).
General procedure. Compound 2a-b (ca. 0.50 mmol) was dissolved in CH2Cl2 (15 mL), and MeI (1.5 equivalents) was added to the solution. The resulting mixture was stirred at room temperature for 2 h, and then charged on an alumina column. Elution with CH2Cl2 allowed to separate impurities, then the fraction corresponding to the product was collected using MeCN/MeOH (95/5 v/v) as eluent. General procedure. Compound 2a-b (ca. 0.50 mmol) was dissolved in CH 2 Cl 2 (15 mL), and MeI (1.5 equivalents) was added to the solution. The resulting mixture was stirred at room temperature for 2 h, and then charged on an alumina column. Elution with CH 2 Cl 2 allowed to separate impurities, then the fraction corresponding to the product was collected using MeCN/MeOH (95/5 v/v) as eluent.

X-Ray Crystallography
Crystal data and collection details for 3 are reported in Table 5. Data were recorded on a Bruker APEX II diffractometer equipped with a PHOTON100 detector using Mo-Kα radiation. The crystal appeared to be non-merohedrally twinned. The program CELL_NOW (G. M. Sheldrick, CELL_NOW, Version 2008/4, 2008 was used in order to determine the two twin domains and their orientation matrices. After integration, data were corrected for Lorentz polarization and absorption effects (empirical absorption correction TWINABS) [70]. The structure was solved by direct methods and refined by full-matrix least-squares based on all data using F 2 [71]. Hydrogen atoms were fixed at calculated positions and refined using a riding model. All non-hydrogen atoms were refined with anisotropic displacement parameters. Refinement was performed using the instruction HKLF 5 in SHELXL and one BASF parameter, which refined as 0.276 (7). Because of the twinning, several restraints were applied during refinement. All the atoms were restrained to have similar thermal parameters (SIMU line in SHLEXL, s.u. 0.01). All C, O, and N atoms were restrained to isotropic like behavior (ISOR line in SHELXL, s.u. 0.01).

X-Ray Crystallography
Crystal data and collection details for 3 are reported in Table 5. Data were recorded on a Bruker APEX II diffractometer equipped with a PHOTON100 detector using Mo-Kα radiation. The crystal appeared to be non-merohedrally twinned. The program CELL_NOW (G. M. Sheldrick, CELL_NOW, Version 2008/4, 2008 was used in order to determine the two twin domains and their orientation matrices. After integration, data were corrected for Lorentz polarization and absorption effects (empirical absorption correction TWINABS) [70]. The structure was solved by direct methods and refined by full-matrix least-squares based on all data using F 2 [71]. Hydrogen atoms were fixed at calculated positions and refined using a riding model. All non-hydrogen atoms were refined with anisotropic displacement parameters. Refinement was performed using the instruction HKLF 5 in SHELXL and one BASF parameter, which refined as 0.276 (7). Because of the twinning, several restraints were applied during refinement. All the atoms were restrained to have similar thermal parameters (SIMU line in SHLEXL, s.u. 0.01). All C, O, and N atoms were restrained to isotropic like behavior (ISOR line in SHELXL, s.u. 0.01).

Computational Studies
The electronic structures of the compounds were optimized using the range-separated ωB97X DFT functional [72][73][74] in combination with Ahlrichs' split-valence-polarized basis set [75]. The C-PCM implicit solvation model was added to ωB97X calculations, considering chloroform as a continuous medium [76,77].
Preliminary optimizations were carried out using the hybrid-GGA EDF2 functional [78] in combination with the 6-31G(d,p) basis set [79]. The stationary points were characterized by IR simulations (harmonic approximation), from which zero-point vibrational energies and thermal corrections (T = 25 • C) were obtained. Simulated IR spectra were used to assign the experimentally observed signals [80]. The software used were Gaussian 09 (Gaussian, Inc: Wallingford, CT, USA) [81] and Spartan '16 [82]. Cartesian coordinates of the DFT-optimized structures are collected in a separated .xyz file.

Stability in Aqueous Solutions
Each compound (ca. 10 mg; 3, 4a-d, 5a-b, 6) was dissolved in DMSO-d 6 (0.4 mL), then the solution was diluted with variable volumes of D 2 O. The resulting solution was kept at 37 • C for 72 h. Subsequent 1 H NMR spectra revealed the presence of the respective starting materials together with minor decomposition products (<10%). NaCl was added in ca. 0.05 M concentration to the solutions containing 4a-d, and the obtained mixtures were kept at 37 • C for 72 h before 1 H NMR spectra were recorded (Table 6). In order to perform tests in contact with a cell culture medium, compounds 4c-d, 5a and 6 (ca. 4 mg) were dissolved in DMSO (ca. 1 mL) in a glass tube, then RPMI-1640 medium with sodium bicarbonate, without l-glutamine and phenol red (ca. 3 mL, Merck), was added. The resulting mixture was maintained at 37 • C for 72 h, then it was allowed to cool to room temperature. Dichloromethane (ca. 4 mL) was added, and the mixture was vigorously shaken. An aliquot of the organic phase was analyzed by IR spectroscopy (Table 6).

Electrochemistry
Cyclic voltammograms were measured under an atmosphere of argon using standard Schlenk techniques with a Palmsens4 potentiostat by working with anhydrous and degassed solutions of acetonitrile (MeCN). MeCN was dried and distilled under Ar from the appropriate drying agent (CaH 2 ), and thoroughly deoxygenated with Ar prior to use. The samples were measured at a concentration of 0.1 M and using 0.1 M NBu 4 PF 6 (Merck) as conductive salt. A glassy carbon electrode was used as working electrode, a coiled platinum wire as counter electrode, and a silver wire as a pseudo-reference electrode. Ferrocene (or decamethylferrocene) was added as an internal standard and all spectra were referenced to the ferrocenium/ferrocene couple (Fc + /Fc).

Cell Culture and Cytotoxicity Studies
Human ovarian carcinoma (A2780 and A2780cisR) cell lines were obtained from the European Collection of Cell Cultures. The human embryonic kidney (HEK-293) cell line was obtained from ATCC (Merck, Buchs, Switzerland). Penicillin streptomycin, RPMI 1640 GlutaMAX (where RPMI = Roswell Park Memorial Institute), and DMEM GlutaMAX media (where DMEM = Dulbecco's modified Eagle medium) were obtained from Life Technologies, and fetal bovine serum (FBS) was obtained from Merck. The cells were cultured in RPMI 1640 GlutaMAX (A2780 and A2780cisR) and DMEM GlutaMAX (HEK-293) media containing 10% heat-inactivated FBS and 1% penicillin streptomycin at 37 • C and CO 2 (5%). The A2780cisR cell line was routinely treated with cisplatin (2 µM) in the media to maintain cisplatin resistance. The cytotoxicity was determined using the 3-(4,5-dimethyl 2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay [83]. Cells were seeded in flat-bottomed 96-well plates as a suspension in a prepared medium (100 µL aliquots and approximately 4300 cells/well) and preincubated for 24 h. Stock solutions of compounds were prepared in DMSO and were diluted in medium. The solutions were sequentially diluted to give a final DMSO concentration of 0.5% and a final compound concentration range (0-200 µM). Cisplatin and RAPTA-C [62] were tested in aqueous solution as a positive (0-100 µM) and negative (200 µM) controls, respectively. The compounds were added to the preincubated 96-well plates in 100 µL aliquots, and the plates were incubated for a further 72 h. MTT (20 µL, 5 mg/mL in Dulbecco's phosphate buffered saline) was added to the cells, and the plates were incubated for a further 4 h. The culture medium was aspirated and the purple formazan crystals, formed by the mitochondrial dehydrogenase activity of vital cells, were dissolved in DMSO (100 µL/well). The absorbance of the resulting solutions, directly proportional to the number of surviving cells, was quantified at 590 nm using a SpectroMax M5e multimode microplate reader (using SoftMax Pro software, version 6.2.2). The percentage of surviving cells was calculated from the absorbance of wells corresponding to the untreated control cells. The reported IC 50 values are based on the means from two independent experiments, each comprising four tests per concentration level.

ROS Production Assessment
The intracellular increase of reactive oxygen species (ROS) upon treatment with the analyzed complexes was measured by using the DCFH-DA (2 ,7 -dichlorodihydrofluorescein diacetate, Merck) assay, based on cellular uptake of the non-fluorescent diacetate following deacetylation by esterases (2 ,7 -dichlorodihydrofluorescein, DCFH) and oxidation to the fluorescent dichlorofluorescein (2 ,7 -dichloro-fluorescein, DCF) [84]. A2780 cells were seeded at concentration of 4300 cells/well/90 µL of complete growth medium into 96-well plates and allowed to proliferate for 24 h. Then cells were treated following the manufacturer's protocol. Briefly, the culture medium was supplemented with 100 µL of the fluorogenic probe solution and cells were incubated under standard tissue culture conditions of 5% CO 2 at 37 • C. After 1 h, the cells were exposed with a final concentration of 10 µM of the tested compounds and maintained at 5% CO 2 at 37 • C; H 2 O 2 100 µM was used as a positive control. Stock solutions of compounds were prepared as described above. Cells incubated with DMSO at a concentration of 0.1% in supplemented RPMI were used as control. The fluorescence was measured over 24 h with an excitation wavelength of 485 nm and with a 535 nm emission filter by Multilabel Counter (PerkinElmer, Waltham, USA). Analysis was conducted in triplicate and experimental data were reported as mean ± SD. Statistical differences were analyzed using one-way analysis of variance (ANOVA) and a Tukey test was used for post hoc analysis. A p-value <0.05 was considered as statistically significant.

Conclusions
The bridging vinyliminium ligand in cationic diiron complexes can be modified by introducing sulphur-or selenium-functions according to well defined regio-and stereoselective reaction pathways. Some of the resulting, air stable diiron and tetrairon compounds display a strong antiproliferative activity against human ovarian carcinoma cell lines, the activity of some compounds on the A2780 cell line being superior than that of cisplatin and substantially maintained on the A2780cisR resistant cell line. Experiments suggest that the chalcogen function (especially the presence of an E-E bridge) is associated with good stability in aqueous media, enhancing interference of compounds with cellular redox processes.
Supplementary Materials: The following are available online: DFT structures, cyclic voltammograms, NMR spectra of products (signals around 0 ppm due to some silicon grease). CCDC reference number 1983327 (3) contains the supplementary crystallographic data for the X-ray studies reported in this paper. These data can be obtained free of charge at www.ccdc.cam.ac.uk/conts/retrieving.html (or from the Cambridge Crystallographic Data Centre, 12, Union Road, Cambridge CB2 1EZ, UK; fax: (internat.) +44-1223/336-033; e-mail: deposit@ccdc.cam.ac.uk).